For Heads-of · Practitioner
Multimodal AI
AI systems that process and combine more than one type of data (text, images, audio, video) to understand and respond to the world.
- multimodal
- vision
- audio
- integration
Capabilities
Multimodal models can answer questions about images, describe video, generate images from text, or move between modalities in ways a single-modality model cannot.
Complexity
Combining modalities makes a model more complex and its behaviour harder to predict or explain. Instructions can also arrive hidden in an image or an audio clip, where text-only filters do not look.
Governance
A multimodal system can inherit risks from every input type, so an image-text model can combine visual and linguistic bias. Image and video capabilities also bring EU AI Act prohibitions closer: inferring sensitive attributes from images is biometric categorisation of sensitive attributes, and reading emotion at work or in education is workplace and education emotion recognition.
Related Terms
Deepfake
AI-generated or manipulated image, audio or video content that resembles real people, objects, places or events and would falsely appear to a person to be authentic. For an organisation, the most immediate risk is impersonation fraud: a cloned voice or a faked video call instructing a payment.
Embedding
A numerical representation of text, images or other data in a high-dimensional space, where items with similar meaning sit close together.
EU AI Act
The European Union's regulation of AI systems by risk category, in force since 1 August 2024 and applying in stages. As of October 2026 the prohibitions, the AI literacy duty, the general-purpose model obligations and the Article 50 transparency duties all apply. The 2026 Digital Omnibus, Regulation (EU) 2026/1744, deferred the high-risk obligations to December 2027 and August 2028, softened the literacy duty and added two prohibitions. It reaches any organisation whose AI output is used in the EU, wherever the organisation sits.
Large Language Model (LLM)
A deep learning model trained on very large amounts of text, able to generate and interpret human-like language across a wide range of topics and tasks.
Transparency
The disclosure of information about an AI system to the people who need it: that AI is being used, what it does, what it is for and what risks it poses, in terms they can understand. For people whose personal data it uses, being informed is also a legal right.
Related risks and controls
Biometric categorisation of sensitive attributes
Image and voice inputs are what make the Article 5 biometric prohibitions reachable in the first place.
Workplace and education emotion recognition
The prohibited use that multimodal input most directly enables.
AI transparency and disclosure labelling
Article 50 labelling applies to generated images, audio and video, not only text.